Re: [HCP-Users] MEG_anatomy_transform.txt
Georgos, I finally have one bit of progress here. The latest FieldTrip installed OK and I was able to use the ft_read_mri function to read the T1 (T1w_acpc_dc_restore.nii) that comes out of the standard HCP pipeline. It showed this transform: -10 0 91 0 1 0 -127 0 0 1 -73 0 0 0 1 Which I presume is the equivalent of the vox2spm transform, one of the two I need for Brainstorm. I am showing my ignorance and inexperience here, but I thought the [load_nii.m] I obtained elsewhere and use to load this T1 would give the same (and complete) contents of the T1.nii file. I also thought that [ft_read_mri.m] would show the exact same contents as the load_nii.m script. But such is not the case. Different fields are output, though I can see the contents of the transform in the load_nii version in the s_row_x,y,z vectors. I still need the second transform, vox2bti, which in my case would be something like vox2EGI for the EGI Net Station electrode coordinates, expressed in some head coordinate system. In the HCP-MEG pipeline the vox2bti transform must be created by registering the BTI MEG sensor coordinates to the voxels, which also must utilize the AC location that is somehow marked on the T1. So it seems to me if I could find out where in the HCP-MEG pipeline this is done I could try to recreate that for me EEG case by substituting the electrode coordinates. Can you guide me to someone who might be able to advise me on this aspect of the HCP-MEG pipeline? Thanks, -Jeff From: Michalareas, Giorgos [mailto:giorgos.michalar...@esi-frankfurt.de] Sent: Tuesday, February 14, 2017 6:00 PM To: K Jeffrey Eriksen Cc: hcp-users@humanconnectome.org Subject: RE: [HCP-Users] MEG_anatomy_transform.txt Hi , try using the latest fieldtrip version from https://github.com/fieldtrip/fieldtrip maybe it will solve the issue you have. The T1 file should have a transformation matrix converting voxel indices to actual spatial coordinates. Maybe you can ask the person operating the MR where the T1 comes from Best Giorgos From: K Jeffrey Eriksen [eriks...@ohsu.edu] Sent: Wednesday, February 15, 2017 1:58 AM To: Michalareas, Giorgos Cc: hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org> Subject: RE: [HCP-Users] MEG_anatomy_transform.txt Hi again, Back again to what you wrote: ** The transformation matrix that goes into field transform.vox07mm2spm can be read from the processed MRI anatomical file , i.e. '175237_MEG_anatomy_anatomical.nii' ** There is no nii file as the above to match the txt file of the same name. The only nii file I get in the example is this: T1_acpc_dc_restore.nii.gz I have opened this with (load_nii.m) and it has no transforms in it. So I remain confused. (I have tried to use ft_read_mri but it is missing some dlls and I have asked FieldTrip support about this. But I do not believe there are any transforms in the T1 file I am given) -Jeff From: Michalareas, Giorgos [mailto:giorgos.michalar...@esi-frankfurt.de] Sent: Tuesday, February 14, 2017 3:03 PM To: K Jeffrey Eriksen Cc: hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org> Subject: RE: [HCP-Users] MEG_anatomy_transform.txt Hi Jeff, there no need to run the HCP-MEG pipeline on your anatomical file. I am not sure f you can even do this if you also dont have the MEG head positioning information. You can use the fieldtrip function ft_read_mri to read the MRI file you have. This function outputs a structure and one of the fields is the transformation matrix you are looking for. Best Giorgos From: K Jeffrey Eriksen [eriks...@ohsu.edu] Sent: Tuesday, February 14, 2017 9:52 PM To: Michalareas, Giorgos Cc: hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org> Subject: RE: [HCP-Users] MEG_anatomy_transform.txt Georgos, Thank you for continuing to help me on this. You indicate that 'transform.vox07mm2spm' can be read from the '_MEG_anatomy_anatomical.nii' file. Trouble is, I do not have any such equivalent files for the data sets I work with, as they are not run through the HCP-MEG pipeline. So I do not see how this is relevant to my case. What I need is an HCP-EEG pipeline. My thinking right now is that if I could understand the anatomy part of the HCP-MEG pipeline I could apply that part to my generic HCP files to get this transform. It just seems to be a registration with an SMP atlas of some sort, correct? Could you help me find out where this happens in the HCP-MEG pipeline, or direct me to people who have a good understanding of it? Regarding the 'spm to bti' transform, I think that is
Re: [HCP-Users] MEG_anatomy_transform.txt
Hi ,
try using the latest fieldtrip version from
https://github.com/fieldtrip/fieldtrip
maybe it will solve the issue you have.
The T1 file should have a transformation matrix converting voxel indices to
actual spatial coordinates.
Maybe you can ask the person operating the MR where the T1 comes from
Best
Giorgos
From: K Jeffrey Eriksen [eriks...@ohsu.edu]
Sent: Wednesday, February 15, 2017 1:58 AM
To: Michalareas, Giorgos
Cc: hcp-users@humanconnectome.org
Subject: RE: [HCP-Users] MEG_anatomy_transform.txt
Hi again,
Back again to what you wrote:
**
The transformation matrix that goes into field
transform.vox07mm2spm
can be read from the processed MRI anatomical file , i.e.
'175237_MEG_anatomy_anatomical.nii'
**
There is no nii file as the above to match the txt file of the same name. The
only nii file I get in the example is this:
T1_acpc_dc_restore.nii.gz
I have opened this with (load_nii.m) and it has no transforms in it. So I
remain confused.
(I have tried to use ft_read_mri but it is missing some dlls and I have asked
FieldTrip support about this. But I do not believe there are any transforms in
the T1 file I am given)
-Jeff
From: Michalareas, Giorgos [mailto:giorgos.michalar...@esi-frankfurt.de]
Sent: Tuesday, February 14, 2017 3:03 PM
To: K Jeffrey Eriksen
Cc: hcp-users@humanconnectome.org
Subject: RE: [HCP-Users] MEG_anatomy_transform.txt
Hi Jeff,
there no need to run the HCP-MEG pipeline on your anatomical file. I am not
sure f you can even do this if you also dont have the MEG head positioning
information.
You can use the fieldtrip function ft_read_mri to read the MRI file you have.
This function outputs a structure and one of the fields is the transformation
matrix you are looking for.
Best
Giorgos
From: K Jeffrey Eriksen [eriks...@ohsu.edu]
Sent: Tuesday, February 14, 2017 9:52 PM
To: Michalareas, Giorgos
Cc: hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>
Subject: RE: [HCP-Users] MEG_anatomy_transform.txt
Georgos,
Thank you for continuing to help me on this.
You indicate that ‘transform.vox07mm2spm’ can be read from the
'_MEG_anatomy_anatomical.nii' file. Trouble is, I do not have any such
equivalent files for the data sets I work with, as they are not run through the
HCP-MEG pipeline. So I do not see how this is relevant to my case. What I need
is an HCP-EEG pipeline.
My thinking right now is that if I could understand the anatomy part of the
HCP-MEG pipeline I could apply that part to my generic HCP files to get this
transform. It just seems to be a registration with an SMP atlas of some sort,
correct? Could you help me find out where this happens in the HCP-MEG pipeline,
or direct me to people who have a good understanding of it?
Regarding the ‘spm to bti’ transform, I think that is mainly to register the
MEG sensors to everything else. In my case the positons of the EEG sensors are
given in native MRI space. I will ask Brainstorm people for more help with this
part as I think they should definitely know how to calculate this.
-Jeff
From: Georgios Michalareas [mailto:giorgos.michalar...@esi-frankfurt.de]
Sent: Tuesday, February 14, 2017 12:56 AM
To: K Jeffrey Eriksen
Cc: hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>
Subject: Re: [HCP-Users] MEG_anatomy_transform.txt
Hi Jeffey,
regarding:
" So only two of the transforms in the transform.vox_filename =
'_MEG_anatomy_anatomical.txt' are needed by BS. I just have to figure
out how to create them for my non-MEG data. "
The transformation matrix that goes into field
transform.vox07mm2spm
can be read from the processed MRI anatomical file , i.e.
'175237_MEG_anatomy_anatomical.nii'
by using the function ft_read_mri from fieldtrip toolbox.
As you don't have any MEG coordinate system and only the MRI coordinate system
then there is no transformation from "spm" to "bti". They are the same
coordinate system in your case. In the example from BS they used the
transform.txt file from a subject with MEG. In this case there was a different
MEG coordinate system that is why the two transformation matrices were not
identical.
the simplest way to create your own transform.txt file is by using the function
hcp_write_ascii from the MEG pipeline code that you can download in
ConnectomeDB("megconnectome").
transform =[];
transform.vox07mm2spm = [
-0.7 0 0 90.7
0 0.7 0 -126.7
0 0 0.7 -72.7
0 0 0 1
]; % This is the transformation matrix you read with ft_read_mri from the
processed MRI file
transform.vox07mm2bti =vox07mm2spm;
transformfile=_MEG_anatomy_transform.txt
hcp_write_ascii (transformfile,'transform');
Best
Giorgos
On 2/14/2017 12:41 AM, K Jeffrey Eriksen wrote:
Georgos,
Understood about the
Re: [HCP-Users] MEG_anatomy_transform.txt
Hi again,
Back again to what you wrote:
**
The transformation matrix that goes into field
transform.vox07mm2spm
can be read from the processed MRI anatomical file , i.e.
'175237_MEG_anatomy_anatomical.nii'
**
There is no nii file as the above to match the txt file of the same name. The
only nii file I get in the example is this:
T1_acpc_dc_restore.nii.gz
I have opened this with (load_nii.m) and it has no transforms in it. So I
remain confused.
(I have tried to use ft_read_mri but it is missing some dlls and I have asked
FieldTrip support about this. But I do not believe there are any transforms in
the T1 file I am given)
-Jeff
From: Michalareas, Giorgos [mailto:giorgos.michalar...@esi-frankfurt.de]
Sent: Tuesday, February 14, 2017 3:03 PM
To: K Jeffrey Eriksen
Cc: hcp-users@humanconnectome.org
Subject: RE: [HCP-Users] MEG_anatomy_transform.txt
Hi Jeff,
there no need to run the HCP-MEG pipeline on your anatomical file. I am not
sure f you can even do this if you also dont have the MEG head positioning
information.
You can use the fieldtrip function ft_read_mri to read the MRI file you have.
This function outputs a structure and one of the fields is the transformation
matrix you are looking for.
Best
Giorgos
From: K Jeffrey Eriksen [eriks...@ohsu.edu]
Sent: Tuesday, February 14, 2017 9:52 PM
To: Michalareas, Giorgos
Cc: hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>
Subject: RE: [HCP-Users] MEG_anatomy_transform.txt
Georgos,
Thank you for continuing to help me on this.
You indicate that 'transform.vox07mm2spm' can be read from the
'_MEG_anatomy_anatomical.nii' file. Trouble is, I do not have any such
equivalent files for the data sets I work with, as they are not run through the
HCP-MEG pipeline. So I do not see how this is relevant to my case. What I need
is an HCP-EEG pipeline.
My thinking right now is that if I could understand the anatomy part of the
HCP-MEG pipeline I could apply that part to my generic HCP files to get this
transform. It just seems to be a registration with an SMP atlas of some sort,
correct? Could you help me find out where this happens in the HCP-MEG pipeline,
or direct me to people who have a good understanding of it?
Regarding the 'spm to bti' transform, I think that is mainly to register the
MEG sensors to everything else. In my case the positons of the EEG sensors are
given in native MRI space. I will ask Brainstorm people for more help with this
part as I think they should definitely know how to calculate this.
-Jeff
From: Georgios Michalareas [mailto:giorgos.michalar...@esi-frankfurt.de]
Sent: Tuesday, February 14, 2017 12:56 AM
To: K Jeffrey Eriksen
Cc: hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>
Subject: Re: [HCP-Users] MEG_anatomy_transform.txt
Hi Jeffey,
regarding:
" So only two of the transforms in the transform.vox_filename =
'_MEG_anatomy_anatomical.txt' are needed by BS. I just have to figure
out how to create them for my non-MEG data. "
The transformation matrix that goes into field
transform.vox07mm2spm
can be read from the processed MRI anatomical file , i.e.
'175237_MEG_anatomy_anatomical.nii'
by using the function ft_read_mri from fieldtrip toolbox.
As you don't have any MEG coordinate system and only the MRI coordinate system
then there is no transformation from "spm" to "bti". They are the same
coordinate system in your case. In the example from BS they used the
transform.txt file from a subject with MEG. In this case there was a different
MEG coordinate system that is why the two transformation matrices were not
identical.
the simplest way to create your own transform.txt file is by using the function
hcp_write_ascii from the MEG pipeline code that you can download in
ConnectomeDB("megconnectome").
transform =[];
transform.vox07mm2spm = [
-0.7 0 0 90.7
0 0.7 0 -126.7
0 0 0.7 -72.7
0 0 0 1
]; % This is the transformation matrix you read with ft_read_mri from the
processed MRI file
transform.vox07mm2bti =vox07mm2spm;
transformfile=_MEG_anatomy_transform.txt
hcp_write_ascii (transformfile,'transform');
Best
Giorgos
On 2/14/2017 12:41 AM, K Jeffrey Eriksen wrote:
Georgos,
Understood about the naming.
Next question: I have found the Brainstorm code that reads the transform file:
%% = MRI=>MNI TRANSFORMATION =
% Convert transformations from "Brainstorm MRI" to "FieldTrip voxel"
Tbst2ft = [diag([-1, 1, 1] ./ sMri.Voxsize), [size(sMri.Cube,1); 0; 0]; 0 0 0
1];
% Set the MNI=>SCS transformation in the MRI
Tmni = transform.vox07mm2spm * Tbst2ft;
sMri.NCS.R = Tmni(1:3,1:3);
sMri.NCS.T = Tmni(1:3,4);
% MNI coordinates for the AC/PC/IH fiducials
AC = [0, 3, -4] ./ 1000;
PC = [0, -25, -2] ./ 1000;
Re: [HCP-Users] MEG_anatomy_transform.txt
Never mind, I see on the FT site that all the MEX files may not be available
for all platforms. I need to compile them for my particular combo of OS+Matlab.
I will investigate this with local help.
-Jeff
From: hcp-users-boun...@humanconnectome.org
[mailto:hcp-users-boun...@humanconnectome.org] On Behalf Of K Jeffrey Eriksen
Sent: Tuesday, February 14, 2017 3:42 PM
To: Michalareas, Giorgos
Cc: hcp-users@humanconnectome.org
Subject: Re: [HCP-Users] MEG_anatomy_transform.txt
Giorgos,
I have installed the FieldTrip toolbox associated with megconnectome 3.0, that
is fieldtrip-r10442. I set the path as per the instructions on the FT site, and
ran the ft_defaults. When I try to call ft_read_mri, I get this:
filename = T1w_acpc_dc_restore.nii
>> [mri]=ft_read_mri(filename);
Invalid MEX-file
'J:\Projects\HCP_MEG\fieldtrip-r10442\fileio\private\ft_getopt.mexw64':
The specified procedure could not be found..
Error in ft_read_mri (line 79)
mriformat = ft_getopt(varargin, 'dataformat');
I suppose I have not set up the MEX-files properly, but I do not know how to do
that.
-Jeff
From: Michalareas, Giorgos [mailto:giorgos.michalar...@esi-frankfurt.de]
Sent: Tuesday, February 14, 2017 3:03 PM
To: K Jeffrey Eriksen
Cc: hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>
Subject: RE: [HCP-Users] MEG_anatomy_transform.txt
Hi Jeff,
there no need to run the HCP-MEG pipeline on your anatomical file. I am not
sure f you can even do this if you also dont have the MEG head positioning
information.
You can use the fieldtrip function ft_read_mri to read the MRI file you have.
This function outputs a structure and one of the fields is the transformation
matrix you are looking for.
Best
Giorgos
From: K Jeffrey Eriksen [eriks...@ohsu.edu]
Sent: Tuesday, February 14, 2017 9:52 PM
To: Michalareas, Giorgos
Cc: hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>
Subject: RE: [HCP-Users] MEG_anatomy_transform.txt
Georgos,
Thank you for continuing to help me on this.
You indicate that 'transform.vox07mm2spm' can be read from the
'_MEG_anatomy_anatomical.nii' file. Trouble is, I do not have any such
equivalent files for the data sets I work with, as they are not run through the
HCP-MEG pipeline. So I do not see how this is relevant to my case. What I need
is an HCP-EEG pipeline.
My thinking right now is that if I could understand the anatomy part of the
HCP-MEG pipeline I could apply that part to my generic HCP files to get this
transform. It just seems to be a registration with an SMP atlas of some sort,
correct? Could you help me find out where this happens in the HCP-MEG pipeline,
or direct me to people who have a good understanding of it?
Regarding the 'spm to bti' transform, I think that is mainly to register the
MEG sensors to everything else. In my case the positons of the EEG sensors are
given in native MRI space. I will ask Brainstorm people for more help with this
part as I think they should definitely know how to calculate this.
-Jeff
From: Georgios Michalareas [mailto:giorgos.michalar...@esi-frankfurt.de]
Sent: Tuesday, February 14, 2017 12:56 AM
To: K Jeffrey Eriksen
Cc: hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>
Subject: Re: [HCP-Users] MEG_anatomy_transform.txt
Hi Jeffey,
regarding:
" So only two of the transforms in the transform.vox_filename =
'_MEG_anatomy_anatomical.txt' are needed by BS. I just have to figure
out how to create them for my non-MEG data. "
The transformation matrix that goes into field
transform.vox07mm2spm
can be read from the processed MRI anatomical file , i.e.
'175237_MEG_anatomy_anatomical.nii'
by using the function ft_read_mri from fieldtrip toolbox.
As you don't have any MEG coordinate system and only the MRI coordinate system
then there is no transformation from "spm" to "bti". They are the same
coordinate system in your case. In the example from BS they used the
transform.txt file from a subject with MEG. In this case there was a different
MEG coordinate system that is why the two transformation matrices were not
identical.
the simplest way to create your own transform.txt file is by using the function
hcp_write_ascii from the MEG pipeline code that you can download in
ConnectomeDB("megconnectome").
transform =[];
transform.vox07mm2spm = [
-0.7 0 0 90.7
0 0.7 0 -126.7
0 0 0.7 -72.7
0 0 0 1
]; % This is the transformation matrix you read with ft_read_mri from the
processed MRI file
transform.vox07mm2bti =vox07mm2spm;
transformfile=_MEG_anatomy_transform.txt
hcp_write_ascii (transformfile,'transform');
Best
Giorgos
On 2/14/2017 12:41 AM, K Jeffrey Eriksen wrote:
Georgos,
Understood about the naming.
Next question: I have found the Brainstorm code that reads the transform file:
%% = MRI=>MNI TRANSFORMATION =
% Convert transformations from "Brainstorm MRI" to "F
Re: [HCP-Users] MEG_anatomy_transform.txt
Giorgos,
I have installed the FieldTrip toolbox associated with megconnectome 3.0, that
is fieldtrip-r10442. I set the path as per the instructions on the FT site, and
ran the ft_defaults. When I try to call ft_read_mri, I get this:
filename = T1w_acpc_dc_restore.nii
>> [mri]=ft_read_mri(filename);
Invalid MEX-file
'J:\Projects\HCP_MEG\fieldtrip-r10442\fileio\private\ft_getopt.mexw64':
The specified procedure could not be found..
Error in ft_read_mri (line 79)
mriformat = ft_getopt(varargin, 'dataformat');
I suppose I have not set up the MEX-files properly, but I do not know how to do
that.
-Jeff
From: Michalareas, Giorgos [mailto:giorgos.michalar...@esi-frankfurt.de]
Sent: Tuesday, February 14, 2017 3:03 PM
To: K Jeffrey Eriksen
Cc: hcp-users@humanconnectome.org
Subject: RE: [HCP-Users] MEG_anatomy_transform.txt
Hi Jeff,
there no need to run the HCP-MEG pipeline on your anatomical file. I am not
sure f you can even do this if you also dont have the MEG head positioning
information.
You can use the fieldtrip function ft_read_mri to read the MRI file you have.
This function outputs a structure and one of the fields is the transformation
matrix you are looking for.
Best
Giorgos
From: K Jeffrey Eriksen [eriks...@ohsu.edu]
Sent: Tuesday, February 14, 2017 9:52 PM
To: Michalareas, Giorgos
Cc: hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>
Subject: RE: [HCP-Users] MEG_anatomy_transform.txt
Georgos,
Thank you for continuing to help me on this.
You indicate that 'transform.vox07mm2spm' can be read from the
'_MEG_anatomy_anatomical.nii' file. Trouble is, I do not have any such
equivalent files for the data sets I work with, as they are not run through the
HCP-MEG pipeline. So I do not see how this is relevant to my case. What I need
is an HCP-EEG pipeline.
My thinking right now is that if I could understand the anatomy part of the
HCP-MEG pipeline I could apply that part to my generic HCP files to get this
transform. It just seems to be a registration with an SMP atlas of some sort,
correct? Could you help me find out where this happens in the HCP-MEG pipeline,
or direct me to people who have a good understanding of it?
Regarding the 'spm to bti' transform, I think that is mainly to register the
MEG sensors to everything else. In my case the positons of the EEG sensors are
given in native MRI space. I will ask Brainstorm people for more help with this
part as I think they should definitely know how to calculate this.
-Jeff
From: Georgios Michalareas [mailto:giorgos.michalar...@esi-frankfurt.de]
Sent: Tuesday, February 14, 2017 12:56 AM
To: K Jeffrey Eriksen
Cc: hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org>
Subject: Re: [HCP-Users] MEG_anatomy_transform.txt
Hi Jeffey,
regarding:
" So only two of the transforms in the transform.vox_filename =
'_MEG_anatomy_anatomical.txt' are needed by BS. I just have to figure
out how to create them for my non-MEG data. "
The transformation matrix that goes into field
transform.vox07mm2spm
can be read from the processed MRI anatomical file , i.e.
'175237_MEG_anatomy_anatomical.nii'
by using the function ft_read_mri from fieldtrip toolbox.
As you don't have any MEG coordinate system and only the MRI coordinate system
then there is no transformation from "spm" to "bti". They are the same
coordinate system in your case. In the example from BS they used the
transform.txt file from a subject with MEG. In this case there was a different
MEG coordinate system that is why the two transformation matrices were not
identical.
the simplest way to create your own transform.txt file is by using the function
hcp_write_ascii from the MEG pipeline code that you can download in
ConnectomeDB("megconnectome").
transform =[];
transform.vox07mm2spm = [
-0.7 0 0 90.7
0 0.7 0 -126.7
0 0 0.7 -72.7
0 0 0 1
]; % This is the transformation matrix you read with ft_read_mri from the
processed MRI file
transform.vox07mm2bti =vox07mm2spm;
transformfile=_MEG_anatomy_transform.txt
hcp_write_ascii (transformfile,'transform');
Best
Giorgos
On 2/14/2017 12:41 AM, K Jeffrey Eriksen wrote:
Georgos,
Understood about the naming.
Next question: I have found the Brainstorm code that reads the transform file:
%% = MRI=>MNI TRANSFORMATION =
% Convert transformations from "Brainstorm MRI" to "FieldTrip voxel"
Tbst2ft = [diag([-1, 1, 1] ./ sMri.Voxsize), [size(sMri.Cube,1); 0; 0]; 0 0 0
1];
% Set the MNI=>SCS transformation in the MRI
Tmni = transform.vox07mm2spm * Tbst2ft;
sMri.NCS.R = Tmni(1:3,1:3);
sMri.NCS.T = Tmni(1:3,4);
% MNI coordinates for the AC/PC/IH fiducials
AC = [0, 3, -4] ./ 1000;
PC = [0, -25, -2] ./ 1000;
IH = [0, -10, 60] ./ 1000;
Origin = [0, 0, 0];
% Convert: MNI (meters) => MRI (millimeters)
sMri.NCS.AC = cs_convert(sMri, 'mni', 'mri', AC) .* 1000;
sMri.NCS.PC = cs_
Re: [HCP-Users] MEG_anatomy_transform.txt
Georgos,
Thank you for continuing to help me on this.
You indicate that 'transform.vox07mm2spm' can be read from the
'_MEG_anatomy_anatomical.nii' file. Trouble is, I do not have any such
equivalent files for the data sets I work with, as they are not run through the
HCP-MEG pipeline. So I do not see how this is relevant to my case. What I need
is an HCP-EEG pipeline.
My thinking right now is that if I could understand the anatomy part of the
HCP-MEG pipeline I could apply that part to my generic HCP files to get this
transform. It just seems to be a registration with an SMP atlas of some sort,
correct? Could you help me find out where this happens in the HCP-MEG pipeline,
or direct me to people who have a good understanding of it?
Regarding the 'spm to bti' transform, I think that is mainly to register the
MEG sensors to everything else. In my case the positons of the EEG sensors are
given in native MRI space. I will ask Brainstorm people for more help with this
part as I think they should definitely know how to calculate this.
-Jeff
From: Georgios Michalareas [mailto:giorgos.michalar...@esi-frankfurt.de]
Sent: Tuesday, February 14, 2017 12:56 AM
To: K Jeffrey Eriksen
Cc: hcp-users@humanconnectome.org
Subject: Re: [HCP-Users] MEG_anatomy_transform.txt
Hi Jeffey,
regarding:
" So only two of the transforms in the transform.vox_filename =
'_MEG_anatomy_anatomical.txt' are needed by BS. I just have to figure
out how to create them for my non-MEG data. "
The transformation matrix that goes into field
transform.vox07mm2spm
can be read from the processed MRI anatomical file , i.e.
'175237_MEG_anatomy_anatomical.nii'
by using the function ft_read_mri from fieldtrip toolbox.
As you don't have any MEG coordinate system and only the MRI coordinate system
then there is no transformation from "spm" to "bti". They are the same
coordinate system in your case. In the example from BS they used the
transform.txt file from a subject with MEG. In this case there was a different
MEG coordinate system that is why the two transformation matrices were not
identical.
the simplest way to create your own transform.txt file is by using the function
hcp_write_ascii from the MEG pipeline code that you can download in
ConnectomeDB("megconnectome").
transform =[];
transform.vox07mm2spm = [
-0.7 0 0 90.7
0 0.7 0 -126.7
0 0 0.7 -72.7
0 0 0 1
]; % This is the transformation matrix you read with ft_read_mri from the
processed MRI file
transform.vox07mm2bti =vox07mm2spm;
transformfile=_MEG_anatomy_transform.txt
hcp_write_ascii (transformfile,'transform');
Best
Giorgos
On 2/14/2017 12:41 AM, K Jeffrey Eriksen wrote:
Georgos,
Understood about the naming.
Next question: I have found the Brainstorm code that reads the transform file:
%% = MRI=>MNI TRANSFORMATION =
% Convert transformations from "Brainstorm MRI" to "FieldTrip voxel"
Tbst2ft = [diag([-1, 1, 1] ./ sMri.Voxsize), [size(sMri.Cube,1); 0; 0]; 0 0 0
1];
% Set the MNI=>SCS transformation in the MRI
Tmni = transform.vox07mm2spm * Tbst2ft;
sMri.NCS.R = Tmni(1:3,1:3);
sMri.NCS.T = Tmni(1:3,4);
% MNI coordinates for the AC/PC/IH fiducials
AC = [0, 3, -4] ./ 1000;
PC = [0, -25, -2] ./ 1000;
IH = [0, -10, 60] ./ 1000;
Origin = [0, 0, 0];
% Convert: MNI (meters) => MRI (millimeters)
sMri.NCS.AC = cs_convert(sMri, 'mni', 'mri', AC) .* 1000;
sMri.NCS.PC = cs_convert(sMri, 'mni', 'mri', PC) .* 1000;
sMri.NCS.IH = cs_convert(sMri, 'mni', 'mri', IH) .* 1000;
sMri.NCS.Origin = cs_convert(sMri, 'mni', 'mri', Origin) .* 1000;
%% = MRI=>SCS TRANSFORMATION =
% Set the MRI=>SCS transformation in the MRI
Tscs = transform.vox07mm2bti * Tbst2ft;
sMri.SCS.R = Tscs(1:3,1:3);
sMri.SCS.T = Tscs(1:3,4);
% Standard positions for the SCS fiducials
NAS = [90, 0, 0] ./ 1000;
LPA = [ 0, 75, 0] ./ 1000;
RPA = [ 0, -75, 0] ./ 1000;
Origin = [0, 0, 0];
% Convert: SCS (meters) => MRI (millimeters)
sMri.SCS.NAS= cs_convert(sMri, 'scs', 'mri', NAS) .* 1000;
sMri.SCS.LPA= cs_convert(sMri, 'scs', 'mri', LPA) .* 1000;
sMri.SCS.RPA= cs_convert(sMri, 'scs', 'mri', RPA) .* 1000;
sMri.SCS.Origin = cs_convert(sMri, 'scs', 'mri', Origin) .* 1000;
% Save MRI structure (with fiducials)
bst_save(BstMriFile, sMri, 'v7');
and here is the relevant contents of the transform file I got with the HCP-MEG
example recommended by BS:
transform.vox07mm_filename = 'T1w_acpc_dc_restore.nii.gz';
transform.vox_filename = '175237_MEG_anatomy_anatomical.nii';
transform.vox07mm2spm = [
-0.7 0 0 90.7
0 0.7 0 -126.7
0 0 0.7 -72.7
0 0 0 1
];
transform.vox07mm2bti = [
0.0242485 0.689697 -0.143363 -99.3102
0.702637 -0.0220904 0.00540799 -88.0016
-5.54434e-05 0.143686 0.684532 -77.7522
0 0 0 1
];
So only two of the transforms in the transform.vox_filename =
'_MEG_anatomy_anatomical.txt' are needed by BS. I just have to figure
out how to create them for my non-MEG data. One dis
Re: [HCP-Users] MEG_anatomy_transform.txt
@esi-frankfurt.de] *Sent:* Friday, February 10, 2017 4:50 PM *To:* K Jeffrey Eriksen *Subject:* RE: [HCP-Users] MEG_anatomy_transform.txt I think this should not be a problem. Just use the T1 you have but make sure the fields names of the "transform" structure remain the same as in the description I sent you because I am pretty confident that Brainstorm is expecting to find these exact filenames, Give it a try snd let me know how it goes Best Giorgos *From:*K Jeffrey Eriksen [eriks...@ohsu.edu] *Sent:* Saturday, February 11, 2017 1:02 AM *To:* Michalareas, Giorgos *Subject:* RE: [HCP-Users] MEG_anatomy_transform.txt Georgios, I have read through this now, and have one follow-up question. While we are using the HCP pipeline, our original T1s are 1.0 mm cubic voxels. Could you please comment on this, or possibly modify what you have written to accommodate this fact? Thanks, -Jeff *From:*hcp-users-boun...@humanconnectome.org <mailto:hcp-users-boun...@humanconnectome.org> [mailto:hcp-users-boun...@humanconnectome.org] *On Behalf Of *Georgios Michalareas *Sent:* Friday, February 10, 2017 3:05 PM *To:* hcp-users@humanconnectome.org <mailto:hcp-users@humanconnectome.org> *Subject:* Re: [HCP-Users] MEG_anatomy_transform.txt Dear Jeff, if I have understood correctly you want to import a structural MRI scan for a given HCP subject into Brainstorm for creating the EEG source model. I guess you follow a procedure similar to http://neuroimage.usc.edu/brainstorm/Tutorials/HCP-MEG and Brainstorm cannot find the MEG/anatomy/*_MEG_anatomy_transform.txt file. Although I am not sure whar Brainstorm reads exactly from the transform file , I can tell you what the transorm file contains. It contains a matlab structure with some fields, which I think you could easily create for subject that have no MEG data. First a quick word on coordinate systems: spm: This is the MNI coordinate system. This is the coordinate system that you 0.7 mm processed structural MRI is defined. bti: This is the coordinate system of the MEG . Of course in your case where there was no MEG recorded , this will be the same as the "spm" coordinate system of the structural MRI. If for your EEG you have digitised the location of the EEG sensors relative to the fiducials then of course you ll need to figure out the transformation matrix between the structural MRI and you EEG reference system. But if this is not the case then your "bti" coordinate system should be the same as the "spm" coordinate system. In this case the identity transformation matrices between this coordinate systems would be spm2bti= [ 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 ]; bti2spm= spm2bti; vox: This is the voxel coordinate system of the structural MRI . The MRI file contains in itself the transformation matrix from voxels to actual coordinates in the "spm". To get this transformation matric in matlab you can use the fieldtrip function "ft_read_mri" Let's assume you have done so and you have found that the transformation matrix from voxel to coordinates is vox07mm2spm= [ 0.0296775 0.00303441 -0.996087 92.0035 -0.965804 -0.245889 -0.0309473 159.445 0.253714 -0.970409 0.00438009 90.3328 0 0 0 1 ]; Important remark: The HCP processed structural MRI has a resolution of 0.7mm. For the MEG processing we use a lower resolution version of the structural MRI . So in the description below whenever you see the notation "vox07" refers to the 0.7mm scan and just "vox" to the lower resolution one. In your case of course these should be the same as you only have the 0.7 mm one. The structure matlab variable is called "transform". I have put the description of the fields in form of code , so potentially you could run it and create the structure for the transform file. Its fields are: *%==* *transform.vox07mm_filename *= 'T1w_acpc_dc_restore.nii.gz'; % This is the name of the HCP processed stuctural MRI with 0.7mm resolution *transform.vox_filename* = transform.vox07mm_filename ; % This is supposed to be the lower resolution version used for MEG but here you assign it also to the 0.7mm *transform.vox07mm2spm* = vox07mm2spm; %This is the transformation matrix from voxel to "spm" coordinates mentioned earlier retrieved from the 0.7mm structural MRI. *transform.spm2vox07mm* = inv(transform.vox2spm); % This is the inverse transformation of the above from "spm" coordinates to voxels. *transform.vox2spm *=transform.vox07mm2spm; %This is the transformation matrix from voxel to "spm" coordinates mentioned earlier . As you only have the 0.7mm structural here is the same also for the lower resolution case. *transform.spm2vox* = inv(transform.vox2spm); % This is the inv
Re: [HCP-Users] MEG_anatomy_transform.txt
Georgos, Understood about the naming. Next question: I have found the Brainstorm code that reads the transform file: %% = MRI=>MNI TRANSFORMATION = % Convert transformations from "Brainstorm MRI" to "FieldTrip voxel" Tbst2ft = [diag([-1, 1, 1] ./ sMri.Voxsize), [size(sMri.Cube,1); 0; 0]; 0 0 0 1]; % Set the MNI=>SCS transformation in the MRI Tmni = transform.vox07mm2spm * Tbst2ft; sMri.NCS.R = Tmni(1:3,1:3); sMri.NCS.T = Tmni(1:3,4); % MNI coordinates for the AC/PC/IH fiducials AC = [0, 3, -4] ./ 1000; PC = [0, -25, -2] ./ 1000; IH = [0, -10, 60] ./ 1000; Origin = [0, 0, 0]; % Convert: MNI (meters) => MRI (millimeters) sMri.NCS.AC = cs_convert(sMri, 'mni', 'mri', AC) .* 1000; sMri.NCS.PC = cs_convert(sMri, 'mni', 'mri', PC) .* 1000; sMri.NCS.IH = cs_convert(sMri, 'mni', 'mri', IH) .* 1000; sMri.NCS.Origin = cs_convert(sMri, 'mni', 'mri', Origin) .* 1000; %% = MRI=>SCS TRANSFORMATION = % Set the MRI=>SCS transformation in the MRI Tscs = transform.vox07mm2bti * Tbst2ft; sMri.SCS.R = Tscs(1:3,1:3); sMri.SCS.T = Tscs(1:3,4); % Standard positions for the SCS fiducials NAS = [90, 0, 0] ./ 1000; LPA = [ 0, 75, 0] ./ 1000; RPA = [ 0, -75, 0] ./ 1000; Origin = [0, 0, 0]; % Convert: SCS (meters) => MRI (millimeters) sMri.SCS.NAS= cs_convert(sMri, 'scs', 'mri', NAS) .* 1000; sMri.SCS.LPA= cs_convert(sMri, 'scs', 'mri', LPA) .* 1000; sMri.SCS.RPA= cs_convert(sMri, 'scs', 'mri', RPA) .* 1000; sMri.SCS.Origin = cs_convert(sMri, 'scs', 'mri', Origin) .* 1000; % Save MRI structure (with fiducials) bst_save(BstMriFile, sMri, 'v7'); and here is the relevant contents of the transform file I got with the HCP-MEG example recommended by BS: transform.vox07mm_filename = 'T1w_acpc_dc_restore.nii.gz'; transform.vox_filename = '175237_MEG_anatomy_anatomical.nii'; transform.vox07mm2spm = [ -0.7 0 0 90.7 0 0.7 0 -126.7 0 0 0.7 -72.7 0 0 0 1 ]; transform.vox07mm2bti = [ 0.0242485 0.689697 -0.143363 -99.3102 0.702637 -0.0220904 0.00540799 -88.0016 -5.54434e-05 0.143686 0.684532 -77.7522 0 0 0 1 ]; So only two of the transforms in the transform.vox_filename = '_MEG_anatomy_anatomical.txt' are needed by BS. I just have to figure out how to create them for my non-MEG data. One disconnect I am having is that below (from your earlier email) you say these two are identical, but BS obviously does not think so as they are obviously different in the above snippet: >>transform.vox07mm2spm = vox07mm2spm; %This is the transformation matrix from >>voxel to "spm" coordinates mentioned earlier retrieved from the 0.7mm >>structural MRI. >>transform.vox07mm2bti = transform.vox07mm2spm ; % As spm and bti are >>identical coordinate system this is just the transformation from voxels to >>"spm" coordinates already created; I do not process MRI data myself, concentrating on the EEG analysis) so do not know how to setup and run the HCP pipeline or the special HCP_MEG pipeline. I did try searching all the HCP_MEG pipeline M-files and found no occurrence of the transforms above, so they must be embedded in the shell scripts or compiled C code somewhere, presumably. Can you comment on the above and perhaps point me in the right direction? One idea I have is to try to run the HCP-MEG pipeline minus the actual MEG portions. Another is to find out if the standard HCP pipeline produces the transforms I need. Thanks, -Jeff From: Michalareas, Giorgos [mailto:giorgos.michalar...@esi-frankfurt.de] Sent: Friday, February 10, 2017 4:50 PM To: K Jeffrey Eriksen Subject: RE: [HCP-Users] MEG_anatomy_transform.txt I think this should not be a problem. Just use the T1 you have but make sure the fields names of the "transform" structure remain the same as in the description I sent you because I am pretty confident that Brainstorm is expecting to find these exact filenames, Give it a try snd let me know how it goes Best Giorgos From: K Jeffrey Eriksen [eriks...@ohsu.edu] Sent: Saturday, February 11, 2017 1:02 AM To: Michalareas, Giorgos Subject: RE: [HCP-Users] MEG_anatomy_transform.txt Georgios, I have read through this now, and have one follow-up question. While we are using the HCP pipeline, our original T1s are 1.0 mm cubic voxels. Could you please comment on this, or possibly modify what you have written to accommodate this fact? Thanks, -Jeff From: hcp-users-boun...@humanconnectome.org<mailto:hcp-users-boun...@humanconnectome.org> [mailto:hcp-users-boun...@humanconnectome.org] On Behalf Of Georgios Michalareas Sent: Friday, February 10, 2017 3:05 PM To: hcp-users@humanconnectome.org<mailto:hcp-users@humanconnectome.org> Subject: Re: [HCP-Users] MEG_anatomy_transform.txt Dear Jeff, if I have understood correctly you want to import a structural MRI scan for a given HCP subject into Brainstorm for creating the
Re: [HCP-Users] MEG_anatomy_transform.txt
Hi Giorgios, I believe you have understood my problem correctly and what you have provided here looks like it is appropriate and complete. I will try to follow your example and see how it goes. I also do not know exactly what Brainstorm needs from the transform file, and am trying to get them to tell me. They have at least pointed me to the Brainstorm M-file that imports the HCP-MEG files, and I might be able to figure that out on my own. Thank you for the quick response, much appreciated. -Jeff From: hcp-users-boun...@humanconnectome.org [mailto:hcp-users-boun...@humanconnectome.org] On Behalf Of Georgios Michalareas Sent: Friday, February 10, 2017 3:05 PM To: hcp-users@humanconnectome.org Subject: Re: [HCP-Users] MEG_anatomy_transform.txt Dear Jeff, if I have understood correctly you want to import a structural MRI scan for a given HCP subject into Brainstorm for creating the EEG source model. I guess you follow a procedure similar to http://neuroimage.usc.edu/brainstorm/Tutorials/HCP-MEG and Brainstorm cannot find the MEG/anatomy/*_MEG_anatomy_transform.txt file. Although I am not sure whar Brainstorm reads exactly from the transform file , I can tell you what the transorm file contains. It contains a matlab structure with some fields, which I think you could easily create for subject that have no MEG data. First a quick word on coordinate systems: spm: This is the MNI coordinate system. This is the coordinate system that you 0.7 mm processed structural MRI is defined. bti: This is the coordinate system of the MEG . Of course in your case where there was no MEG recorded , this will be the same as the "spm" coordinate system of the structural MRI. If for your EEG you have digitised the location of the EEG sensors relative to the fiducials then of course you ll need to figure out the transformation matrix between the structural MRI and you EEG reference system. But if this is not the case then your "bti" coordinate system should be the same as the "spm" coordinate system. In this case the identity transformation matrices between this coordinate systems would be spm2bti= [ 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 ]; bti2spm= spm2bti; vox: This is the voxel coordinate system of the structural MRI . The MRI file contains in itself the transformation matrix from voxels to actual coordinates in the "spm". To get this transformation matric in matlab you can use the fieldtrip function "ft_read_mri" Let's assume you have done so and you have found that the transformation matrix from voxel to coordinates is vox07mm2spm= [ 0.0296775 0.00303441 -0.996087 92.0035 -0.965804 -0.245889 -0.0309473 159.445 0.253714 -0.970409 0.00438009 90.3328 0 0 0 1 ]; Important remark: The HCP processed structural MRI has a resolution of 0.7mm. For the MEG processing we use a lower resolution version of the structural MRI . So in the description below whenever you see the notation "vox07" refers to the 0.7mm scan and just "vox" to the lower resolution one. In your case of course these should be the same as you only have the 0.7 mm one. The structure matlab variable is called "transform". I have put the description of the fields in form of code , so potentially you could run it and create the structure for the transform file. Its fields are: %== transform.vox07mm_filename = 'T1w_acpc_dc_restore.nii.gz'; % This is the name of the HCP processed stuctural MRI with 0.7mm resolution transform.vox_filename = transform.vox07mm_filename ; % This is supposed to be the lower resolution version used for MEG but here you assign it also to the 0.7mm transform.vox07mm2spm = vox07mm2spm; %This is the transformation matrix from voxel to "spm" coordinates mentioned earlier retrieved from the 0.7mm structural MRI. transform.spm2vox07mm = inv(transform.vox2spm); % This is the inverse transformation of the above from "spm" coordinates to voxels. transform.vox2spm =transform.vox07mm2spm; %This is the transformation matrix from voxel to "spm" coordinates mentioned earlier . As you only have the 0.7mm structural here is the same also for the lower resolution case. transform.spm2vox = inv(transform.vox2spm); % This is the inverse transformation of the above from "spm" coordinates to voxels. transform.spm2bti = spm2bti; % This is the transformation between spm and bti coordinate systems. In this case it is assumed they are the same so it the identity matrix created earlier. transform.bti2spm = bti2spm; %Same case here transform.vox07mm2bti = transform.vox07mm2spm ; % As spm and bti are identical coordinate system this is just the transformation from voxels to "spm" coordinates already created; transform.bti2vox07mm = transform.spm2vox07mm; % The same is true for the inverse transformation. tran
Re: [HCP-Users] MEG_anatomy_transform.txt
Dear Jeff, if I have understood correctly you want to import a structural MRI scan for a given HCP subject into Brainstorm for creating the EEG source model. I guess you follow a procedure similar to http://neuroimage.usc.edu/brainstorm/Tutorials/HCP-MEG and Brainstorm cannot find the MEG/anatomy/*_MEG_anatomy_transform.txt file. Although I am not sure whar Brainstorm reads exactly from the transform file , I can tell you what the transorm file contains. It contains a matlab structure with some fields, which I think you could easily create for subject that have no MEG data. First a quick word on coordinate systems: spm: This is the MNI coordinate system. This is the coordinate system that you 0.7 mm processed structural MRI is defined. bti: This is the coordinate system of the MEG . Of course in your case where there was no MEG recorded , this will be the same as the "spm" coordinate system of the structural MRI. If for your EEG you have digitised the location of the EEG sensors relative to the fiducials then of course you ll need to figure out the transformation matrix between the structural MRI and you EEG reference system. But if this is not the case then your "bti" coordinate system should be the same as the "spm" coordinate system. In this case the identity transformation matrices between this coordinate systems would be spm2bti= [ 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 ]; bti2spm= spm2bti; vox: This is the voxel coordinate system of the structural MRI . The MRI file contains in itself the transformation matrix from voxels to actual coordinates in the "spm". To get this transformation matric in matlab you can use the fieldtrip function "ft_read_mri" Let's assume you have done so and you have found that the transformation matrix from voxel to coordinates is vox07mm2spm= [ 0.0296775 0.00303441 -0.996087 92.0035 -0.965804 -0.245889 -0.0309473 159.445 0.253714 -0.970409 0.00438009 90.3328 0 0 0 1 ]; Important remark: The HCP processed structural MRI has a resolution of 0.7mm. For the MEG processing we use a lower resolution version of the structural MRI . So in the description below whenever you see the notation "vox07" refers to the 0.7mm scan and just "vox" to the lower resolution one. In your case of course these should be the same as you only have the 0.7 mm one. The structure matlab variable is called "transform". I have put the description of the fields in form of code , so potentially you could run it and create the structure for the transform file. Its fields are: *%==* *transform.vox07mm_filename *= 'T1w_acpc_dc_restore.nii.gz'; % This is the name of the HCP processed stuctural MRI with 0.7mm resolution *transform.vox_filename* = transform.vox07mm_filename ; % This is supposed to be the lower resolution version used for MEG but here you assign it also to the 0.7mm *transform.vox07mm2spm* = vox07mm2spm; %This is the transformation matrix from voxel to "spm" coordinates mentioned earlier retrieved from the 0.7mm structural MRI. *transform.spm2vox07mm* = inv(transform.vox2spm); % This is the inverse transformation of the above from "spm" coordinates to voxels. *transform.vox2spm *=transform.vox07mm2spm; %This is the transformation matrix from voxel to "spm" coordinates mentioned earlier . As you only have the 0.7mm structural here is the same also for the lower resolution case. *transform.spm2vox* = inv(transform.vox2spm); % This is the inverse transformation of the above from "spm" coordinates to voxels. *transform.spm2bti* = spm2bti; % This is the transformation between spm and bti coordinate systems. In this case it is assumed they are the same so it the identity matrix created earlier. *transform.bti2spm* = bti2spm; %Same case here *transform.vox07mm2bti* = transform.vox07mm2spm ; % As spm and bti are identical coordinate system this is just the transformation from voxels to "spm" coordinates already created; *transform.bti2vox07mm *= transform.spm2vox07mm; % The same is true for the inverse transformation. *transform.vox2bti* = transform.vox2spm; % Same for the lower resolution case used for MEG, which of course yo dont have, as yo uonly have the 0.7 mm so this is the same as for the 0.7mm case *transform.bti2vox *= transform.spm2vox; % Same for the inverse transform of the lower resolution case *transform.vox2bti_registered* = transform.vox2bti; % This is an intermediate transformation from voxel to bti used only for MEG purposed but for your case should be exactly the same as the voxel to bti transformation which of course is the same as the voxel to spm transformation as these coordinate systems are the same in your case. *%==* Once you have created the "transform" variable you have to write it in an ascii file, which is not as strightforward as writing a numberical
